10x spliting barcode caller

.github/workflows/Barcode.Mouse.10x_concat.yml

@@ -0,0 +1,84 @@
+name: Barcode Mouse 10x concatenated reads
+
+on:
+  workflow_dispatch:
+  schedule:
+  - cron: '15 3 * * 2'
+
+env:
+  LAUNCHER: ${{github.workspace}}/isoquant_tests/github/run_barcode_test.py
+  CFG_DIR: /abga/work/andreyp/ci_isoquant/data/barcodes
+  BIN_PATH: /abga/work/andreyp/ci_isoquant/bin/
+  OUTPUT_BASE: /abga/work/andreyp/ci_isoquant/output/${{github.ref_name}}/barcodes/
+
+concurrency:
+  group: ${{github.workflow}}
+  cancel-in-progress: false
+
+jobs:
+  check-changes:
+    runs-on:
+      labels: [isoquant]
+    name: 'Check for recent changes'
+    outputs:
+      has_changes: ${{steps.check.outputs.has_changes}}
+    steps:
+      - name: 'Checkout'
+        uses: actions/checkout@v3
+        with:
+          fetch-depth: 0
+
+      - name: 'Check for commits in last 7 days'
+        id: check
+        run: |
+          # Always run on manual trigger
+          if [ "${{github.event_name}}" = "workflow_dispatch" ]; then
+            echo "has_changes=true" >> $GITHUB_OUTPUT
+            exit 0
+          fi
+          # Check for commits in last 7 days
+          COMMITS=$(git log --oneline --since="7 days ago" | wc -l)
+          if [ "$COMMITS" -gt 0 ]; then
+            echo "Found $COMMITS commits in last 7 days"
+            echo "has_changes=true" >> $GITHUB_OUTPUT
+          else
+            echo "No commits in last 7 days, skipping"
+            echo "has_changes=false" >> $GITHUB_OUTPUT
+          fi
+
+  launch-runner:
+    needs: check-changes
+    if: needs.check-changes.outputs.has_changes == 'true'
+    runs-on:
+      labels: [isoquant]
+    name: 'Running barcode detection QC for 10x concatenated reads'
+
+    steps:
+      - name: 'Cleanup'
+        run: >
+          set -e &&
+          shopt -s dotglob &&
+          rm -rf *
+
+      - name: 'Checkout'
+        uses: actions/checkout@v3
+        with:
+          fetch-depth: 1
+
+      - name: 'Mouse 10x concatenated reads (3M whitelist)'
+        if: always()
+        shell: bash
+        env:
+          STEP_NAME: Mouse.10x_concat.3M.800K
+        run: |
+          export PATH=$PATH:${{env.BIN_PATH}}
+          python3 ${{env.LAUNCHER}} ${{env.CFG_DIR}}/${{env.STEP_NAME}}.yaml -o ${{env.OUTPUT_BASE}}
+
+      - name: 'Mouse 10x concatenated reads (realistic 9K whitelist)'
+        if: always()
+        shell: bash
+        env:
+          STEP_NAME: Mouse.10x_concat.realistic.800K
+        run: |
+          export PATH=$PATH:${{env.BIN_PATH}}
+          python3 ${{env.LAUNCHER}} ${{env.CFG_DIR}}/${{env.STEP_NAME}}.yaml -o ${{env.OUTPUT_BASE}}

isoquant_lib/barcode_calling/__init__.py

@@ -33,6 +33,7 @@
     LinkerBarcodeDetectionResult,
     TSOBarcodeDetectionResult,
     TenXBarcodeDetectionResult,
+    TenXSplitBarcodeDetectionResult,
     SplittingBarcodeDetectionResult,
     ReadStats,
     increase_if_valid,
@@ -49,6 +50,8 @@
     SharedMemoryWrapper,
     TenXBarcodeDetector,
     TenXv2BarcodeDetector,
+    TenXSplittingBarcodeDetector,
+    TenXv2SplittingBarcodeDetector,
     VisiumHDBarcodeDetector,
     UniversalSingleMoleculeExtractor,
     MoleculeStructure
@@ -80,6 +83,7 @@
     'LinkerBarcodeDetectionResult',
     'TSOBarcodeDetectionResult',
     'TenXBarcodeDetectionResult',
+    'TenXSplitBarcodeDetectionResult',
     'SplittingBarcodeDetectionResult',
     'ReadStats',
     'increase_if_valid',
@@ -95,6 +99,8 @@
     # 10x detectors
     'TenXBarcodeDetector',
     'TenXv2BarcodeDetector',
+    'TenXSplittingBarcodeDetector',
+    'TenXv2SplittingBarcodeDetector',
     'VisiumHDBarcodeDetector',
     # universal calling
     'UniversalSingleMoleculeExtractor',

isoquant_lib/barcode_calling/callers/__init__.py

@@ -15,6 +15,7 @@
     LinkerBarcodeDetectionResult: Result for platforms with linker sequences
     TSOBarcodeDetectionResult: Result for Stereo-seq with TSO detection
     TenXBarcodeDetectionResult: Result for 10x Genomics platform
+    TenXSplitBarcodeDetectionResult: Result for 10x split mode (includes TSO position)
     SplittingBarcodeDetectionResult: Result for read splitting modes
     ExtractionResult: Dict-based result for universal molecule extraction
     ReadStats: Statistics tracker for barcode detection
@@ -29,6 +30,8 @@
     SharedMemoryWrapper: Generic shared memory wrapper
     TenXBarcodeDetector: 10x Genomics v3 detector
     TenXv2BarcodeDetector: 10x Genomics v2 detector
+    TenXSplittingBarcodeDetector: 10x v3 splitting detector (concatenated reads)
+    TenXv2SplittingBarcodeDetector: 10x v2 splitting detector (concatenated reads)
     VisiumHDBarcodeDetector: Visium HD detector
     UniversalSingleMoleculeExtractor: Universal barcode detector for custom molecules
 """
@@ -40,6 +43,7 @@
     LinkerBarcodeDetectionResult,
     TSOBarcodeDetectionResult,
     TenXBarcodeDetectionResult,
+    TenXSplitBarcodeDetectionResult,
     SplittingBarcodeDetectionResult,
     ReadStats,
     increase_if_valid,
@@ -66,6 +70,8 @@
 from .tenx import (
     TenXBarcodeDetector,
     TenXv2BarcodeDetector,
+    TenXSplittingBarcodeDetector,
+    TenXv2SplittingBarcodeDetector,
     VisiumHDBarcodeDetector,
 )
 
@@ -81,6 +87,7 @@
     'LinkerBarcodeDetectionResult',
     'TSOBarcodeDetectionResult',
     'TenXBarcodeDetectionResult',
+    'TenXSplitBarcodeDetectionResult',
     'SplittingBarcodeDetectionResult',
     'ExtractionResult',
     'DetectedElement',
@@ -98,6 +105,8 @@
     # 10x detectors
     'TenXBarcodeDetector',
     'TenXv2BarcodeDetector',
+    'TenXSplittingBarcodeDetector',
+    'TenXv2SplittingBarcodeDetector',
     'VisiumHDBarcodeDetector',
     # Universal extraction
     'MoleculeStructure',

isoquant_lib/barcode_calling/callers/base.py

@@ -11,7 +11,7 @@
 """
 
 from collections import defaultdict
-from typing import List, Optional, Dict, Iterable
+from typing import List, Optional, Dict, Iterable, Union
 
 
 def increase_if_valid(val: Optional[int], delta: int) -> Optional[int]:
@@ -253,6 +253,20 @@ def get_additional_attributes(self) -> List[str]:
             attr.append("TSO detected")
         return attr
 
+    def get_fasta_segment_start(self) -> int:
+        """Start position of the FASTA segment to extract for this molecule."""
+        return max(0, self.primer - 25, self.polyT - 75)
+
+    def get_fasta_segment_end(self, seq_len: int) -> int:
+        """End position of the FASTA segment to extract for this molecule."""
+        if self.tso5 == -1:
+            return seq_len
+        return min(seq_len, self.tso5 + 25)
+
+    def get_tso_position(self) -> int:
+        """Return TSO position for use in require_tso checks."""
+        return self.tso5
+
     @staticmethod
     def header() -> str:
         """Static header for class-level access."""
@@ -299,6 +313,43 @@ def header() -> str:
         return BarcodeDetectionResult.header() + "\tpolyT_start\tR1_end"
 
 
+class TenXSplitBarcodeDetectionResult(TenXBarcodeDetectionResult):
+    """Detection result for 10x split mode — includes TSO position for molecule boundary."""
+
+    def __init__(self, read_id: str, barcode: str = BarcodeDetectionResult.NOSEQ,
+                 UMI: str = BarcodeDetectionResult.NOSEQ,
+                 BC_score: int = -1, UMI_good: bool = False, strand: str = ".",
+                 polyT: int = -1, r1: int = -1, tso: int = -1):
+        TenXBarcodeDetectionResult.__init__(self, read_id, barcode, UMI, BC_score, UMI_good, strand, polyT, r1)
+        self.tso: int = tso
+
+    def update_coordinates(self, delta: int) -> None:
+        TenXBarcodeDetectionResult.update_coordinates(self, delta)
+        if self.tso != -1:
+            self.tso += delta
+
+    def get_fasta_segment_start(self) -> int:
+        """Start position of FASTA segment: just before R1 linker."""
+        return max(0, self.r1 - 10) if self.r1 != -1 else 0
+
+    def get_fasta_segment_end(self, seq_len: int) -> int:
+        """End position of FASTA segment: just past the TSO, or end of read."""
+        if self.tso == -1:
+            return seq_len
+        return min(seq_len, self.tso + 35)
+
+    def get_tso_position(self) -> int:
+        """Return TSO position for use in require_tso checks."""
+        return self.tso
+
+    def __str__(self) -> str:
+        return TenXBarcodeDetectionResult.__str__(self) + "\t%d" % self.tso
+
+    @staticmethod
+    def header() -> str:
+        return TenXBarcodeDetectionResult.header() + "\ttso_start"
+
+
 class SplittingBarcodeDetectionResult:
     """Result container for read splitting modes (multiple barcodes per read)."""
 
@@ -307,9 +358,9 @@ class SplittingBarcodeDetectionResult:
     def __init__(self, read_id: str):
         self.read_id: str = read_id
         self.strand: str = "."  # For protocol compatibility
-        self.detected_patterns: List[TSOBarcodeDetectionResult] = []
+        self.detected_patterns: List[Union[TSOBarcodeDetectionResult, TenXSplitBarcodeDetectionResult]] = []
 
-    def append(self, barcode_detection_result: TSOBarcodeDetectionResult) -> None:
+    def append(self, barcode_detection_result: Union[TSOBarcodeDetectionResult, TenXSplitBarcodeDetectionResult]) -> None:
         self.detected_patterns.append(barcode_detection_result)
 
     def empty(self) -> bool: